Multi-analyte proteomic analysis identifies blood-based neuroinflammation, cerebrovascular and synaptic biomarkers in preclinical Alzheimer’s disease

Table 2 Diagnostic accuracy (ROC analysis) of NULISAseq and Simoa biomarkers for Aβ PET positivity

AUC (95% CI)	ROC AUC with Biomarker only			ROC AUC with Biomarker + risk factors
AUC (95% CI)	NULISASeq	Simoa	p-value	NULISAseq	Simoa	p-value
p-tau217	0.905 (0.841–0.969)	0.880 (0.800–0.959)	0.625	0.930 (0.878–0.983)	0.925 (0.874–0.977)	0.891
p-tau231	0.718 (0.615–0.822)	0.632 (0.517–0.746)	0.274	0.806 (0.709–0.903)	0.790 (0.689–0.891)	0.822
p-tau181	0.576 (0.468–0.684)	0.669 (0.560–0.778)	0.237	0.780 (0.678–0.882)	0.793 (0.662–0.893)	0.867
GFAP	0.732 (0.640–0.825)	0.743 (0.653–0.834)	0.867	0.808 (0.717–0.899)	0.816 (0.725–0.908)	0.901
NEFL	0.565 (0.442–0.688)	0.599 (0.480–0.719)	0.695	0.766 (0.655–0.878)	0.791 (0.682–0.901)	0.755
Aβ40	0.539 (0.409–0.669)	0.650 (0.530–0.771)	0.221	0.773 (0.670–0.875)	0.816 (0.719–0.913)	0.547
Aβ42	0.531 (0.411–0.650)	0.550 (0.432–0.667)	0.826	0.779 (0.671–0.886)	0.789 (0.691–0.887)	0.884

Risk factors included age as a numeric variable, sex (male and female), and APOE status (APOE ε4 carrier or non-carrier). The confidence interval of AUC was computed using the bootstrapping approach. P-value was determined using the DeLong test implemented in the pROC package in RStudio

ISSN: 1750-1326